On Fri, 9 Sep 2011, Andy Farnell wrote:
It's more a general theme in the discussion, I think the author's background is web so the casual examples mention it, at least in the bits I browsed - he was talking about erlang as a solution for distributed databases. The main examples for each language seemed standard compsci problems though, sorting, permuting, factoring etc.
But the basis of comparison necessary to justify today's language choices can hardly be made of those standard compsci problems, which are all very small scale.
For example, to see that goto was harmful, you had to see it on a large scale enough that it becomes tangled. If you do it on single-page-or-less compsci problems, goto always seems ok. Many of today's people have learned about goto only by indoctrination, but by doing projects that don't fit on a page (nor on a few), people can learn from experience that goto is harmful.
Multi-processor computing has been a compsci topic for about 30 years already, and distributed computing is nearly as old. TCP/IP-enabled kernels came from Berkeley in 1982, NFS in 1984, X11 officially in 1987 (but the project had been published since early 80's).
« Standard compsci » problems exclude a lot of new things for pedagogical reasons, to stay within the level of difficulty of first-year programming students and middle-year algorithmics students. There isn't a reason to stay within that problem set when the goal is to compare languages for daily use in potentially big practical projects.
| Mathieu Bouchard ---- tél: +1.514.383.3801 ---- Villeray, Montréal, QC